On Non-Uniform Rational B-Splines Surface Neural Networks
Neural Processing Letters
| Hi-index | 0.00 |
This paper presents a new coding algorithm making use of B-spline surfaces. Points on the surface are regarded as image data, and the values of defining polygon vertices are coded. Simulation results show the possibility that the performance of this algorithm becomes better than that of discrete cosine transform (DCT) when parameter values of the surface are set according to the position and the steepness of edges, and the number of polygon vertices is set according to the fineness of the image and the bit rate. It is known that a B-spline surface is transformed by transforming the defining polygon net. Therefore, using this algorithm, it becomes possible to apply image transformations to compressed data (polygon vertices) instead of to decoded data